Debian Bug report logs - #413934
minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Markus Meier <rattynator@googlemail.com>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Thu, 08 Mar 2007 00:14:07 +0100

Package: minicom
Version: 2.2-5
Severity: normal

If minicom is started with LANG=de_DE.UTF-8 and serial port is set wrong
(wrong Bps/Par/Bits), it crashes with the following error message:

xminicom: ../iconv/loop.c:425: internal_utf8_loop_single: Zusicherung
»inptr - bytebuf > (state->__count & 7)« nicht erfüllt.

If I start it with LANG=POSIX it seems to work fine.

-- System Information:
Debian Release: 4.0
  APT prefers testing
  APT policy: (900, 'testing'), (300, 'unstable')
Architecture: amd64 (x86_64)
Shell:  /bin/sh linked to /bin/bash
Kernel: Linux 2.6.17-2-xen-amd64
Locale: LANG=de_DE.UTF-8, LC_CTYPE=de_DE.UTF-8 (charmap=UTF-8)

Versions of packages minicom depends on:
ii  libc6                       2.3.6.ds1-13 GNU C Library: Shared libraries
ii  libncurses5                 5.5-5        Shared libraries for terminal hand

Versions of packages minicom recommends:
ii  lrzsz                        0.12.21-4.1 Tools for zmodem/xmodem/ymodem fil

-- no debconf information

Thanks
Markus

Message #10 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Adam Lackorzynski <adam@os.inf.tu-dresden.de>

To: Markus Meier <rattynator@googlemail.com>, 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Thu, 8 Mar 2007 10:33:40 +0100

Hi,

On Thu Mar 08, 2007 at 00:14:07 +0100, Markus Meier wrote:
> If minicom is started with LANG=de_DE.UTF-8 and serial port is set wrong
> (wrong Bps/Par/Bits), it crashes with the following error message:
> 
> xminicom: ../iconv/loop.c:425: internal_utf8_loop_single: Zusicherung
> »inptr - bytebuf > (state->__count & 7)« nicht erfüllt.

Can you give me a concrete example of the wrong configuration? I can't
trigger this assertion (on ia32).


Thanks,
Adam
-- 
Adam                 adam@os.inf.tu-dresden.de
  Lackorzynski         http://os.inf.tu-dresden.de/~adam/

Message #15 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Markus Meier <rattynator@googlemail.com>

To: Adam Lackorzynski <adam@os.inf.tu-dresden.de>

Cc: 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Thu, 8 Mar 2007 14:04:09 +0100

Am Donnerstag, 8. März 2007 10:33 schrieb Adam Lackorzynski:
> On Thu Mar 08, 2007 at 00:14:07 +0100, Markus Meier wrote:
> > If minicom is started with LANG=de_DE.UTF-8 and serial port is set
> > wrong (wrong Bps/Par/Bits), it crashes with the following error
> > message:
> >
> > xminicom: ../iconv/loop.c:425: internal_utf8_loop_single:
> > Zusicherung »inptr - bytebuf > (state->__count & 7)« nicht erfüllt.
>
> Can you give me a concrete example of the wrong configuration? I
> can't trigger this assertion (on ia32).

I booted pfsense (Firewall derivated from FreeBSD) and tried to watch 
the boot process via serial console. pfsense wants me to set the serial 
port to 9600 8N1, but I firstly forgot this and kept the old setting, 
which was 38400 8N1. I think the resultant garbage on the terminal 
raised the foregoing error. I also did some further investigation and 
tried this on x86 and x86-64, but the error appears on both computers.

I also found the following typing error:
When I leave minicom with strg a + x it says "Modem zurüksetzen" but 
here a 'c' is missing, as it has to be labeled to "Modem zurücksetzen".

Markus

-- 
Just because I'm not paranoid doesn't mean they're not out to get me!

Message #20 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Adam Lackorzynski <adam@os.inf.tu-dresden.de>

To: Markus Meier <rattynator@googlemail.com>

Cc: 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Sat, 10 Mar 2007 17:41:19 +0100

On Thu Mar 08, 2007 at 14:04:09 +0100, Markus Meier wrote:
> Am Donnerstag, 8. März 2007 10:33 schrieb Adam Lackorzynski:
> > On Thu Mar 08, 2007 at 00:14:07 +0100, Markus Meier wrote:
> > > If minicom is started with LANG=de_DE.UTF-8 and serial port is set
> > > wrong (wrong Bps/Par/Bits), it crashes with the following error
> > > message:
> > >
> > > xminicom: ../iconv/loop.c:425: internal_utf8_loop_single:
> > > Zusicherung »inptr - bytebuf > (state->__count & 7)« nicht erfüllt.
> >
> > Can you give me a concrete example of the wrong configuration? I
> > can't trigger this assertion (on ia32).
> 
> I booted pfsense (Firewall derivated from FreeBSD) and tried to watch 
> the boot process via serial console. pfsense wants me to set the serial 
> port to 9600 8N1, but I firstly forgot this and kept the old setting, 
> which was 38400 8N1. I think the resultant garbage on the terminal 
> raised the foregoing error. I also did some further investigation and 
> tried this on x86 and x86-64, but the error appears on both computers.

It's hard for me to do anything about this since this is an assertion
from inside glibc. To start, I probably need to reproduce this...

 
> I also found the following typing error:
> When I leave minicom with strg a + x it says "Modem zurüksetzen" but 
> here a 'c' is missing, as it has to be labeled to "Modem zurücksetzen".

Thanks, this one is fixed in upstream.



Adam
-- 
Adam                 adam@os.inf.tu-dresden.de
  Lackorzynski         http://os.inf.tu-dresden.de/~adam/

Message #25 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Mike Crowe <mac@mcrowe.com>

To: adam@os.inf.tu-dresden.de

Cc: 413934@bugs.debian.org, Markus Meier <rattynator@googlemail.com>

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Mon, 26 Mar 2007 16:38:18 +0100

[Message part 1 (text/plain, inline)]

Hi Adam & Markus,

I think I may have found a recipe for reproducing the minicom assert
with LANG=en_GB.UTF-8. As Markus reported the problem goes away with
LANG=posix.

Arrange for Minicom receive the following (invalid UTF-8) byte
sequence over the serial port:

f8 e2 82 ac c2 a3 0a

The error message I see is:

minicom: ../iconv/loop.c:425: internal_utf8_loop_single: Assertion `inptr - bytebuf > (state->__count & 7)' failed.

I've attached a file containing the bytes too.

Mike.

[crash.cap (text/plain, attachment)]

Message #30 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Marcus Better <marcus@better.se>

To: 413934@bugs.debian.org

Subject: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Wed, 11 Apr 2007 15:46:29 +0200

[Message part 1 (text/plain, inline)]

I just got the same crash when receiving some random garbage data from an 
embedded device. The locale is sv_SE.UTF-8.

[Message part 2 (application/pgp-signature, inline)]

Message #33 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Martin Godisch <martin@godisch.de>

To: Adam Lackorzynski <adam@os.inf.tu-dresden.de>, 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Sun, 15 Apr 2007 08:10:07 +0200

[Message part 1 (text/plain, inline)]

On Sat, Mar 10, 2007 at 17:41:19 +0100, Adam Lackorzynski wrote:

> > > > If minicom is started with LANG=de_DE.UTF-8 and serial port is set
> > > > wrong (wrong Bps/Par/Bits), it crashes with the following error
> > > > message:
> > > >
> > > > xminicom: ../iconv/loop.c:425: internal_utf8_loop_single:
> > > > Zusicherung »inptr - bytebuf > (state->__count & 7)« nicht erfüllt.

> It's hard for me to do anything about this since this is an assertion
> from inside glibc. To start, I probably need to reproduce this...

Does a strace log help? The bug can be reproduced by pointing minicom to
a fifo (with empty scriptprog, minit, mreset) and putting Mike Crowe's
crash.cap into it.

Martin

[minicom.log (text/plain, attachment)]

Message #36 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Martin Godisch <martin@godisch.de>

To: Adam Lackorzynski <adam@os.inf.tu-dresden.de>, 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Sun, 15 Apr 2007 08:15:58 +0200

On Sun, Apr 15, 2007 at 08:10:07 +0200, Martin Godisch wrote:

> Does a strace log help? The bug can be reproduced by pointing minicom to
> a fifo (with empty scriptprog, minit, mreset) and putting Mike Crowe's
> crash.cap into it.

s/strace log/backtrace/

Message #41 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Adam Lackorzynski <adam@os.inf.tu-dresden.de>

To: 413934@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Sun, 15 Apr 2007 12:15:37 +0200

On Sun Apr 15, 2007 at 08:10:07 +0200, Martin Godisch wrote:
> On Sat, Mar 10, 2007 at 17:41:19 +0100, Adam Lackorzynski wrote:
> 
> > > > > If minicom is started with LANG=de_DE.UTF-8 and serial port is set
> > > > > wrong (wrong Bps/Par/Bits), it crashes with the following error
> > > > > message:
> > > > >
> > > > > xminicom: ../iconv/loop.c:425: internal_utf8_loop_single:
> > > > > Zusicherung »inptr - bytebuf > (state->__count & 7)« nicht erfüllt.
> 
> > It's hard for me to do anything about this since this is an assertion
> > from inside glibc. To start, I probably need to reproduce this...
> 
> Does a strace log help? The bug can be reproduced by pointing minicom to
> a fifo (with empty scriptprog, minit, mreset) and putting Mike Crowe's
> crash.cap into it.

The backtrace does not really help. I'm not sure what to do as this
looks like a glibc issue. On the other side there's #115566.
Reassign/reopen?

This little patch makes minicom crash instantly when started, when using
LANG=en_GB.UTF-8 (or another utf8 locale).

Unfortunately creating a little test program which triggers this
assertion does not seem so easy.

This issue also puts the question whether it is good to display (serial)
input with the local locale or rather use "C"? Or something
configurable? A work around for this could also be to set the locale to
"C" when outputting input data.

Any opinions?

Index: main.c
===================================================================
RCS file: /cvsroot/minicom/minicom/src/main.c,v
retrieving revision 1.16
diff -u -r1.16 main.c
--- main.c	31 Oct 2005 12:13:51 -0000	1.16
+++ main.c	15 Apr 2007 10:11:06 -0000
@@ -681,8 +681,21 @@
     /* Update the timer. */
     timer_update();
 
-    /* Check for I/O or timer. */
-    x = check_io(portfd_connected, 0, 1000, buf, &blen);
+    {
+      static int foo = 0;
+      if (!foo) {
+	x = 1;
+	buf[0] = 248;
+	buf[1] = 226;
+	buf[2] = 130;
+	buf[3] = 172;
+	buf[4] = 194;
+	buf[5] = 163;
+	blen = 6;
+	foo = 1;
+      } else
+        x = check_io(portfd_connected, 0, 1000, buf, &blen);
+    }
 
     /*  Send data from the modem to the screen. */
     if ((x & 1) == 1) {



Adam
-- 
Adam                 adam@os.inf.tu-dresden.de
  Lackorzynski         http://os.inf.tu-dresden.de/~adam/

Message #46 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Martin Godisch <martin@godisch.de>

To: 413934@bugs.debian.org, libc6@packages.debian.org

Cc: control@bugs.debian.org

Subject: Re: Bug#413934: minicom crashes if LANG=de_DE.UTF-8 is set - with LANG=POSIX it works fine

Date: Sun, 9 Dec 2007 11:38:59 +0100

reassign 413934 libc6
thanks

Hello libc6-maintainers,

would you have a look into this bugs #413934 and #423017, please?
minicom crashes due to a libc6-internal assertion which (I think)
should not happen in the first way...

Thanks! -- Martin

Message #55 received at 413934@bugs.debian.org (full text, mbox, reply):

From: Mike Crowe <mac@mcrowe.com>

To: 413934@bugs.debian.org

Subject: Bug appears to be fixed in squeeze glibc

Date: Thu, 29 Jul 2010 12:30:28 +0100

I've tested minicom in a squeeze chroot using the byte sequence I
provided earlier. I could not reproduce the problem. Minicom displays
a strange character (as would be expected) but does not assert.

I tested with:

ii  libc6                              2.11.2-2                           Embedded GNU C Library: Shared libraries
ii  minicom                            2.4-2                              friendly menu driven serial communication program

So, I think we can call this fixed in squeeze.

Thanks.

Mike.

Message #60 received at 413934-done@bugs.debian.org (full text, mbox, reply):

From: Aurelien Jarno <aurelien@aurel32.net>

To: Mike Crowe <mac@mcrowe.com>, 413934-done@bugs.debian.org

Subject: Re: Bug#413934: Bug appears to be fixed in squeeze glibc

Date: Sat, 31 Jul 2010 05:58:04 +0200

Version: 2.11.2-2

On Thu, Jul 29, 2010 at 12:30:28PM +0100, Mike Crowe wrote:
> I've tested minicom in a squeeze chroot using the byte sequence I
> provided earlier. I could not reproduce the problem. Minicom displays
> a strange character (as would be expected) but does not assert.
> 
> I tested with:
> 
> ii  libc6                              2.11.2-2                           Embedded GNU C Library: Shared libraries
> ii  minicom                            2.4-2                              friendly menu driven serial communication program
> 
> So, I think we can call this fixed in squeeze.
> 

Thanks for the follow-up, I am marking the bug as fixed in version
2.11.2-2.

-- 
Aurelien Jarno	                        GPG: 1024D/F1BCDB73
aurelien@aurel32.net                 http://www.aurel32.net

Send a report that this bug log contains spam.